Business Process Quality Metrics: Log-Based Complexity of Workflow Patterns
We believe that analysis tools for BPM should provide other analytical capabilities besides verification. Namely, they should provide mechanisms to analyze the complexity of workflows. High complexity in workflows may result in poor understandability, err
- PDF / 205,428 Bytes
- 8 Pages / 430 x 660 pts Page_size
- 51 Downloads / 198 Views
Abstract. We believe that analysis tools for BPM should provide other analytical capabilities besides verification. Namely, they should provide mechanisms to analyze the complexity of workflows. High complexity in workflows may result in poor understandability, errors, defects, and exceptions leading processes to need more time to develop, test, and maintain. Therefore, excessive complexity should be avoided. The major goal of this paper is to describe a quality metric to analyze the complexity of workflow patterns from a log-based perspective. Keywords: workflow, process log, workflow complexity, business process quality metrics, business process analysis.
1 Introduction Workflow verification tools such as Woflan [1] are indispensable for the current generation of WfMS. Yet, another desirable category of tools that allows building better workflows are tools that implement workflow quality metrics. In the area of software engineering, quality metrics have shown their importance for good programming practices and software designs. Since there are strong similarities between software programs and business process designs, several researchers have recognized the potential of quality metrics in business process management [2-5]. In [6], Vanderfeesten et al. suggest that quality metrics to analyze business processes can be classified into four distinct categories: coupling, cohesion, complexity, modularity and size. In this paper we focus our attention on developing quality metrics to evaluate the complexity of workflow models [7]. Workflow complexity should not be confused with algorithmic complexity measures (e.g. Big-Oh “O”-Notation), whose aim is to compare the performance of algorithms [7]. Workflow complexity can be defined as the degree to which a workflow is difficult to analyze, understand or explain. It can be characterized by the number and intricacy of task interfaces, transitions, conditional and parallel branches, the existence of loops, roles, task categories, the types of data structures, and other workflow characteristics. ∗
This work was partially funded by FCT, POCTI-219, and FEDER.
R. Meersman and Z. Tari et al. (Eds.): OTM 2007, Part I, LNCS 4803, pp. 427–434, 2007. © Springer-Verlag Berlin Heidelberg 2007
428
J. Cardoso
In this paper, we present a metric to calculate the Log-Based Complexity (LBC) of workflow patterns [8]. Since our analysis of complexity is based on flow descriptions, we devise complexity metrics for each workflow pattern. The idea of this metric is to relate complexity with the number of different log traces that can be generated from the execution of a workflow. If a workflow always generates the same entries (i.e., the same task ID) in the process log then its complexity is minimal. On the other hand, if a workflow can generate n! distinct log entries (where n is the number of tasks of a workflow) then its complexity is higher. This paper is structured as follows. The second section presents the related work. In section 3, a new complexity metric for workflow patterns is p
Data Loading...
